Debby,

Thanks - I had already shortened both the quantum and timeslice on the
DS and ES queues, and I'm now waiting for the thrashing to begin <g>.

When you say a CS - and I assume that you mean a higher priority
process - can't preempt a DS as often under 5.0 and later, I assume you
mean that the lower priority process will be allowed to achieve its
quantum.  Under what conditions will such preemption occur now?

I also was unaware of the default DS timeslice value change.  We
typically kept ours at 400 and hadn't reset to default since the 5.0
update.

Lee


______________________________ Reply Separator _________________________________
Subject: Re: Performance question (CS session running DS process)
Author:  DebbyG <[log in to unmask]> at ~INTERNET
Date:    10/24/96 1:51 AM


Lee, you didn't say what version of MPE/iX you were running on, so here's
a generalization.  As of 5.0, the situations where a CS process can
preempt a DS process are much fewer.  It's one of the reasons why the
timeslice for the DS queue was shortened by default to 200 instead of the
2000 it had been at on 4.0 and earlier.  I don't remember the new
conditions offhand -- my notes are back at work.  Basically though, the
old trick of putting intensive processes in the DS queue doesn't work as
well on 5.0 because the CS processes typically can't preempt the DS
process.  I'd check the quantum and timeslice on the DS queue to make sure
it isn't too long.

Good luck!

Debby ;)

####################################################################
Debby Gray
[log in to unmask]  (work)
  Sunnyvale, CA                             [log in to unmask]
       (personal)
Me?  Opinions?  If no one else will claim them, neither will I...
####################################################################